摘要
针对异常数据的不利影响和类数难以判别的问题提出了一种带约束的最小离差平方和系统聚类法,包括对初始数据进行线性变换、聚类检验指标分析、主因素分析等。通过对实际调查样本数据进行聚类,挖掘和分析客户群中所存在的不同特征的组群,得到了直观的聚类过程和较合理的分组结果。
To the questions of adverse effect of unusual data and difficulties in defining the number of clusters, a method of Restricted Minimum Variance Hierarchical Cluster(RMVHC) was proposed, including varying initial data linearly, inspection standard analysis and main factors analysis etc. Through clustering to the sample data of investigation, excavating and analysing various characteristics existing in customers actually, cluster course and relatively rational grouping result were directly perceived through the senses.
出处
《计算机应用》
CSCD
北大核心
2005年第1期45-48,共4页
journal of Computer Applications
关键词
数据挖掘
聚类分析
客户关系管理
Data Mining
clustering
customer relationship management
